December 2009 - Boost-users

Binary (non-character-based) file iterators
by Jason Felice 24 Dec '09

24 Dec '09

I'm looking for binary (value_type = unsigned char) input and output iterators for files. This is as opposed to a character base iterator (which rules out ostream_iterator<>). I don't need anything fancy (in fact, I'm just copying whole files to and from archives). I know that Sprit has an input iterator which is a forward iterator, and I could use that. I've already started to write them with Boost.Iterator; but this seems like something you should expect to find somewhere in boost already.

4 5

[serialization] extended_type_info has no member named get_debug_info
by Patrick Horgan 24 Dec '09

24 Dec '09

Building my own code against trunk I started getting this error. Anyone know off the top of your head what it is before I dig it out? archive/detail/oserializer.hpp:398:36: error: ‘const class boost::serialization::extended_type_info’ has no member named ‘get_debug_info’

2 1

[Iterators] Filtering and conversion
by Ruediger Berlich 24 Dec '09

24 Dec '09

Hi there, I'm playing with the iterators library at the moment. Say I have a std::vector<boost::shared_ptr<base> > , with base being the base class of derived1 and derived2. Pointers to derived1 and derived2 are stored in arbitrary order in the vector. Would it be possible to create an iterator that filters out the derived1 objects (including implicit conversion and possibly dereferencing of the resulting shared_ptr<derivedx>) ? So what I'm looking for is something like this: /*******************************************************/ main() { vector<base> x; for (int i=0; i<10; i++) { if(i%2==0) x.push_back(shared_ptr<derived1>(new derived1(i))); else x.push_back(shared_ptr<derived2>(new derived2(i))); } conversion_iterator<derived1> it(x.end()); // ++it will skip over derived2 objects and set itself // to x.end() if no derived1 is remaining for(it=x.begin(); it!=x.end(); ++it) { // Implies implicit conversion and dereferencing // of shared_ptr<base> it->someDerived1SpecificCall(); } } /*******************************************************/ The filter_iterator already does part of what I'd want here, but cannot help with the conversion part. I'm not asking for a solution to the above problem, just whether you think this can be done at all. I had the impression that the iterators library assumes that the types of the sequence being iterated over and the type being returned are identical. Also, while it=x.begin() can be handled by a conversion constructor. it!=x.end() does not appear to be accessible with the predefined comparison operators. In any case thanks and have a nice Christmas time, Ruediger

2 1

Re: [Boost-users] mpl/fusion: metafunction to 'rangify'
by Hicham Mouline 23 Dec '09

23 Dec '09

>> typedef >> boost::fusion::result_of::as_vector<boost::fusion::result_of::transform< >> params,make_fixed_or_range>::type>::type >> params_fixed_or_range; >> >> -Christopher >It does. >Of course, the return type of fusion::transform contains the compile-time >information required. >Thanks very much, >Now I'll look at spirit (and some magic code to generate the Qi parser for >this new sequence) to fill it up with values from an input stream. I have simplified the code to http://codepad.org/NDM2Ub7z I have a couple of needs now: 1. attempt to transform pfr back to params. This should succeed if all elements of the fusion sequence are fixed values and throw an exception if any of them is a range. Should I write this as a free floating function, as a conversion operator? 2. there will be many params types, params_1 .... params_n, each of those structs will yield after a fusion::transform the sequences params_fixed_of_range_1... params_fixed_of_range_n. I need the following behavior (I would write for this a free floating template function, templated on the sequence type : one of the params_fixed_of_range_i): template <typename params_fixed_of_range> void do_work(const params_fixed_of_range& seq_i) { /// /// generate a inner for loop for all the ranges in seq_i /// for e.g., if seq_i contains /// { d1= 5.; d2= 4.0..5.0..0.1; d3=-1.5..1.5..0.6; i4=4..26..1; i5=4; i6=0; i7=0; } /// then I would want to generate the following code, or if impossible /// maybe by runtime recursion /// for (double d2=4.0; d2<=5.0; d2+=0.1) for (double d3=-1.5; d3<=1.5; d3+=0.6) for (unsigned int i4=4; i4<=26; i4+=1) { // I know d1=5. i5=4 i6=0 i7=0 // construct a params frm current d1,d2,d3,i4,i5,i6,i7 // do some work based on params // return value, this will be used to plot graphs } }; This seems very doable with boost.fusion, and extremely complicated and not flexible without it. I realize there are a lot of questions into 1 email but appreciate the help you can give me, Best regards,

1 0

Possibly Memory Leaostks in Boost.Date_Time (Boost 1.41) ?
by Daniel Vallejos 23 Dec '09

23 Dec '09

Hello! Memory Leaks in Boost.Date_Time (Boost 1.41) ? Code: #include <boost/date_time.hpp> #include <iostream> using namespace boost::posix_time; using namespace std; int main() { time_duration interval = seconds(120); cout << "interval = " << interval << endl; } Valgrind profile: $ valgrind ./date_time_test ==8666== Memcheck, a memory error detector ==8666== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al. ==8666== Using Valgrind-3.5.0-Debian and LibVEX; rerun with -h for copyright info ==8666== Command: ./date_time_test ==8666== interval = 00:02:00 ==8666== ==8666== HEAP SUMMARY: ==8666== in use at exit: 1,595 bytes in 27 blocks ==8666== total heap usage: 48 allocs, 21 frees, 5,650 bytes allocated ==8666== ==8666== LEAK SUMMARY: ==8666== definitely lost: 0 bytes in 0 blocks ==8666== indirectly lost: 0 bytes in 0 blocks ==8666== possibly lost: 619 bytes in 20 blocks ==8666== still reachable: 976 bytes in 7 blocks ==8666== suppressed: 0 bytes in 0 blocks Environment: $ uname -a Linux electroforce 2.6.31-15-generic #50-Ubuntu SMP Tue Nov 10 14:53:52 UTC 2009 x86_64 GNU/Linux $ g++ --version g++ (Ubuntu 4.4.2-5ubuntu1) 4.4.2

3 2

Boost.Units - Can't figure out how to use it
by Brandon Olivares 23 Dec '09

23 Dec '09

Hi, I've been reading the documentation for Boost.Units. I understand most of what it is saying, but am still confused on how to do many things with this library. Most of the examples are using that test_system.hpp. Does this mean that you actually have to explicitly define the units you want to use? I am confused because I see the header files for the units I want to use, e.g., light_year or astronomical_unit, but I can't figure out how to actually use them. If I use boost::units::astronomical::light_second, it cannot find it. I am able to use the SI units just fine, such as: Quantity<si::length> l = 4.0 * si::meters; But beyond that, I'm lost. I also don't know which headers I should include. I don't know if there's one or two headers that include all the basics I would need, or if I have to include each of them individually. Here's a simple program I was trying to write just to test light_second, but do not know how to make it work or which headers to include: int main() { using namespace boost::units; using namespace boost::units::astronomical; using namespace std; quantity<light_second> l1 = 4.0 * light_second; cout << "l1 = " << l1 << endl; return 0; } I'm using Boost 1.41 and I'm on Windows using Cygwin. Thanks, Brandon -- www.creatorspiritus.com Blog and forum discussing living a Christ-centered life Follow me on Twitter: http://twitter.com/devbanana Follow creatorspiritus.com on Twitter: http://twitter.com/creatorspiritus

2 5

bjam? cmake?
by Roman Neuhauser 23 Dec '09

23 Dec '09

Hello, can someone explain to me what's been going on with Boost.Build? I cannot find a conclusive reference in the mailing lists, but it looks like bjam is being replaced with cmake. Is that right? If it is, how far has the effort gotten? I have a few tiny utilities written in C++ currently built and installed with GNU make, which I'd like to replace with something more portable, and I don't want to learn bjam and bbv2 if their replacement is around the corner. TIA -- Roman Neuhauser

3 4

[date_time] Delimiters are ignored
by Vadim Guchenko 23 Dec '09

23 Dec '09

Hello, All! Why the following code works while input string has wrong delimiters and too large value of seconds? #include <boost/date_time/posix_time/posix_time.hpp> #include <iostream> #include <sstream> #include <string> #include <locale> int main() { std::string str("08a02b61 15c10d2008"); std::istringstream is(str); is.imbue(std::locale(is.getloc(), new boost::posix_time::time_input_facet( "%H:%M:%S %d/%m/%Y"))); boost::posix_time::ptime ptime; is >> ptime; if (is) std::cout << ptime << "\n"; else std::cout << "Error\n"; } It outputs: 2008-Oct-15 08:03:01 -- Best regards, Vadim Guchenko [yhw at relost dot net].

1 0

iterator adaptor (complex->real)
by Neal Becker 23 Dec '09

23 Dec '09

I've been using an iterator adaptor to reinterpret between vectors of scalar and vectors of complex. Problem is, it relies on real(z) returning an lvalue. This is true in current gnu libstdc++, but looks like it's not in the current standard, and appears to be going away (real(z) will return rvalue). I need some help fixing this code in my complex_to_real iterator adaptor: template<typename BaseIterator> class const_c_to_r_adapt : public boost::iterator_adaptor< const_c_to_r_adapt<BaseIterator>, BaseIterator, typename std::iterator_traits<BaseIterator>::value_type::value_type, use_default, typename std::iterator_traits<BaseIterator>::value_type::value_type ... scalar_type & dereference() const { return (cnt % 2 == 0) ? real (const_cast<complex_type&>(*this- >base_reference())) : imag (const_cast<complex_type&>(*this- >base_reference())); } Changing it to return an rvalue is not acceptable. I need the functionality of returning an lvalue. Any ideas?

1 0

mpl/fusion: metafunction to 'rangify'
by Hicham Mouline 23 Dec '09

23 Dec '09

I am making small progress in my attempt to generate a new mpl sequence that I will use to read from a stream. The following code fails to compile: #include <boost/tuple/tuple.hpp> #include <boost/mpl/vector_c.hpp> #include <boost/mpl/transform.hpp> #include <boost/fusion/adapted/struct/adapt_struct.hpp> #include <boost/fusion/include/adapt_struct.hpp> #include <boost/fusion/mpl.hpp> // makes params a MPL sequence #include <boost/fusion/support/is_sequence.hpp> #include <boost/fusion/include/is_sequence.hpp> struct params { double d1; double d2; double d3; unsigned int i4; unsigned int i5; unsigned int i6; unsigned int i7; }; BOOST_FUSION_ADAPT_STRUCT( params, (double, d1) (double, d2) (double, d3) (unsigned int, i4) (unsigned int, i5) (unsigned int, i6) (unsigned int, i7) ) BOOST_MPL_ASSERT(( boost::fusion::traits::is_sequence<params> )); // passes template <typename T1> struct rangify { typedef boost::tuple<T1, T1, T1> type; /// min max incr }; template <typename T> struct fixed_or_range { typedef struct { union { T fixed; typename rangify<T>::type range; }; bool is_fixed; } type; }; typedef boost::mpl::transform< params, fixed_or_range<boost::mpl::_> >::type params_fixed_or_range; I understand that after the macro BOOST_FUSION_ADAPT_STRUCT , params is a fusion sequence. Including <boost/fusion/mpl.hpp> made params a MPL sequence, and therefore I can call boost::mpl::transform<> on it. The compilation error is c:\program files (x86)\boost_1_41_0\boost\mpl\clear.hpp(30) : error C2903: 'apply' : symbol is neither a class template nor a function template 1> c:\program files (x86)\boost_1_41_0\boost\mpl\transform.hpp(113) : see reference to class template instantiation 'boost::mpl::clear<Sequence>' being compiled 1> with 1> [ 1> Sequence=params 1> ] 1> c:\program files (x86)\boost_1_41_0\boost\mpl\eval_if.hpp(41) : see reference to class template instantiation 'boost::mpl::transform1<P1,P2,P3>' being compiled 1> with 1> [ 1> P1=params, 1> P2=fixed_or_range<boost::mpl::_1>, 1> P3=boost::mpl::na 1> ] 1> c:\program files (x86)\boost_1_41_0\boost\mpl\transform.hpp(138) : see reference to class template instantiation 'boost::mpl::eval_if<C,F1,F2>' being compiled 1> with 1> [ 1> C=boost::mpl::or_<boost::mpl::is_na<boost::mpl::na>,boost::mpl::is_lambda_ex pression<fixed_or_range<boost::mpl::_1>>,boost::mpl::not_<boost::mpl::is_seq uence<fixed_or_range<boost::mpl::_1>>>>, 1> F1=boost::mpl::transform1<params,fixed_or_range<boost::mpl::_1>,boost::mpl:: na>, 1> F2=boost::mpl::transform2<params,fixed_or_range<boost::mpl::_1>,boost::mpl:: na,boost::mpl::na> 1> ] 1> c:\users\hich\documents\visual studio 2008\projects\test\test1\main.cpp(51) : see reference to class template instantiation 'boost::mpl::transform<Seq1,Seq2OrOperation>' being compiled 1> with 1> [ 1> Seq1=params, 1> Seq2OrOperation=fixed_or_range<boost::mpl::_1> 1> ] Regards,

2 4